Two Tandem RNase III Cleavage Sites Determine betT mRNA Stability in Response to Osmotic Stress in Escherichia coli

While identifying genes regulated by ribonuclease III (RNase III) in Escherichia coli, we observed that steady-state levels of betT mRNA, which encodes a transporter mediating the influx of choline, are dependent on cellular concentrations of RNase III. In the present study, we also observed that steady-state levels of betT mRNA are dependent on RNase III activity upon exposure to osmotic stress, indicating the presence of cis-acting elements controlled by RNase III in betT mRNA. Primer extension analyses of betT mRNA revealed two tandem RNase III cleavage sites in its stem-loop region, which were biochemically confirmed via in vitro cleavage assays. Analyses of cleavage sites suggested the stochastic selection of cleavage sites by RNase III, and mutational analyses indicated that RNase III cleavage at either site individually is insufficient for efficient betT mRNA degradation. In addition, both the half-life and abundance of betT mRNA were significantly increased in association with decreased RNase III activity under hyper-osmotic stress conditions. Our findings demonstrate that betT mRNA stability is controlled by RNase III at the post-transcriptional level under conditions of osmotic stress.


Introduction
RNase III plays an important role in regulating gene expression by degrading and processing mRNA in both prokaryotes and eukaryotes [1,2]. The RNase III family enzymes specifically cleave double-stranded RNA (dsRNA), creating 59-phosphate and 39hydroxyl termini with a two-nucleotide overhang. In Escherichia coli, RNase III, encoded by the rnc gene, consists of a ribonuclease domain (amino acid residues 21-149) and a dsRNA-binding domain (residues 155-209). E. coli RNase III functions as a homodimer in which two ribonuclease domains form a single processing center, and each domain contributes to the cleavage of one RNA strand of the duplex substrate [3,4].
Several E. coli mRNA transcripts have been identified whose abundance is dependent on the cellular concentration and activity of RNase III, including rnc [5], pnp [6], bdm [7], corA [8], proU [9], mltD [10], and rng [11] mRNA. A recent genome-wide analysis of such RNase III concentration-dependent transcripts indicated that the stability of a large number of mRNA transcripts can be controlled by RNase III [7]. This study further identified a subgroup of mRNA transcripts encoding factors associated with the osmosensing of E. coli that are downregulated by RNase III. Among them, the mechanisms of RNase III-mediated regulation of bdm [7] and proU [9] genes have been identified, which revealed that RNase III activity on bdm and proU mRNA is significantly altered in E. coli subjected to hyper-and hypo-osmotic stress.
These results indicate that RNase III activity is regulated by external signals in E. coli. However, betT, another important osmoregulator, has not been studied with respect to RNase IIImediated regulation upon osmotic stress although betT mRNA abundance was found to be dependent on cellular RNase III concentrations.
The betT gene forms part of the osmoregulatory system Bet regulon, which participates in the synthesis of glycine betaine from externally supplied choline. The Bet system is composed of the following: BetA, a choline dehydrogenase; BetB, a betaine aldehyde dehydrogenase; BetI, a regulatory protein; and BetT, a choline transporter [12][13][14][15]. BetT protein belongs to the betainecholine-carinitine transporter family and has been characterized as a ubiquitous proton-coupled transporter for choline. The functional role of BetT in the regulation of osmolarity is well studied. BetT mainly transports external choline into cells with high affinity (Km = 8 mM) when the external concentration of choline is low [16]. At high choline concentrations (Km = 1.5 mM), choline uptake is mediated by the ProU transport system [17]. However, the functional correlation between the Bet regulon and the ProU transport system in the regulation of cellular osmolarity is unclear [16][17][18].
The physiological role of BetT in osmotic stress resistance has been previously reported [16]; however, the regulation mechanisms of betT gene expression and the factors involved have not been fully characterized. In the present study, we investigated the effects of RNase III activity on betT gene expression in E. coli. Our results provide direct evidence that RNase III controls the degradation of betT mRNA by cleaving at two sites in its open reading frame region, thereby suggesting a physiological relationship between the regulation of RNase III activity and osmotic stress resistance.

Strains and plasmids
Construction of the E. coli strain MG1655rnc has been previously described [8]. To overexpress betT, a DNA fragment encoding betT was amplified using the primers betT-59 (59-ATGCGGCCGCGAATTTGATTTTAAATAGTA, in which the transcriptional initiation site of the 59-betT gene is underlined) and betT-39 (59-ATGCGGCCGCTCACGCGTCCGGGAA-CATCA, in which the complementary sequence of the translational stop codon of the betT open reading frame is underlined), and cloned into the NotI site in pCAT924 [7,19,20]. The resulting plasmid (pBetRS1) overexpresses betT under the control of a constitutive trp c promoter. To express betT mRNA containing a single nucleotide substitution at the RNase III cleavage sites (C33U and C39U), pBetRS1-C33U and pBetRS1-C39U were constructed as follows: DNA fragments were amplified using overlap extension polymerase chain reaction (PCR) method and cloned into the NotI site in pCAT924. The primers used were betT-59, betT-48F (59-GGTGTTTTACACCTCCGCCG), betT-1610R (59-TAATTCATCAGGCGCGAGAG), and betT-11D-67R (59-CGGCGGAGG TGTAAAACACCACCGGATT-GATTTTATCCTTTTCCCTGCTGTGTGAAAG, in which the mutated nucleotide is underlined) for pBetRS1-C33U, and betT-59, betT-48F, betT-1610R, and betT-13I-67R (59-CGGC-GGAGGTGTAAAACACCACCGGATTAATTTTGTCCTTT-TCCCTGCTG, in which the mutated nucleotide is underlined) for pBetRS1-C39U. The MG1655betT, MG1655proV, and MG1655betTproV strains were constructed via one-step inactivation of the chromosomal gene by using the method described by Datsenko and Wanner [21]. The primers used to construct betTdeleted strains were betT-H1 (59-CATATGCAGACATGGCG-CGGTTTTATGCAATAACAAGTGTAGGCTGGAGCTGC-TTC, in which the complementary a 36 nt-long sequence of the betT 39-untranslated region (UTR) is underlined) and betT-H2 (59-GATTTTAAATAGTAACAATAACAGTGGGGATACTGGC-ATATGAATATCCTCCTTA, in which a 36 nt-long sequence of the betT 59-UTR is underlined) with pKD3 as a template. The primers proV-59-UTR-Km (59-TATCGACATAGACAAATAAA-GGAATCTTTCTATTGCAGAGCGCTTTTGAAGCTCAC, in which the sequence of the proV 59-UTR is underlined) and proV-39-UTR-Km (59-TGGCGTGGTATCCCACGGATTATTTT-GATCAGCCATCCCTTATTAGAAGAACTCGT, in which the complementary sequence of the proV 39-UTR and the coding region is underlined) were used to amplify the kanamycin antibiotic resistance marker in pKD13 for construction of proV-deleted strains.

In vitro cleavage analysis
His-tagged RNase III purification and cleavage assays were performed as previously described [1]. Template DNA fragments for the synthesis of full-length betT mRNA and the model hairpin RNA, which spans nucleotides 21-95 in the betT coding region (the first nucleotide of the start codon is at position 1) and encompasses the RNase III cleavage sites, were obtained by amplifying the corresponding sequence in pBetRS1 using the following primers: T7-betT-F (59-TAATACGACTCACTATAG-GAATTTGATTTTAAATAGTA, in which the 59-UTR of betT is underlined) and betT-R (59-TCACGCGTCCGGGAACATCA, in which the complementary sequence of the stop codon is underlined) for a full-length betT transcript, as well as T7-betT-21F (59-TAATACGACTCACTATAGCAGGGAAAAGGACAAAA-TCAA, in which the 59 region of model hairpin RNA is underlined) and betT-95R (59-GTTGTCAGGGAAAACAA-CAAA, in which the complementary sequence of the 59 region of model hairpin RNA is underlined) for the model hairpin RNA. Labeled RNA (0.2 pmol) was incubated with 1 ng of purified RNase III in the presence of 0.25 mg ml 21 of yeast tRNA (Ambion) and 20 U of RNase inhibitor (Takara) in 20 ml of cleavage buffer (30 mM Tris-HCl, pH 7.9, 160 mM NaCl, 0.1 mM dithiothreitol, 0.1 mM ethylenediamine tetraacetic acid (EDTA), pH 8.0). Cleavage reactions were initiated by adding 10 mM MgCl 2 at 37uC. Samples were removed at various time intervals (0, 1, 2, and 4 min) and mixed with an equal volume of gel-loading buffer (deionized-formamide 95%, EDTA (pH 8.0), 0.025% sodium dodecyl sulfate, 0.025% xylene cyanol, and 0.025% bromophenol blue). The samples were denatured at 65uC for 10 min and separated on an 8% or 10% polyacrylamide gel containing 8 M urea and 0.56Tris-borate EDTA.

Primer extension
Primer extension analysis was performed using total RNA purified via heated phenol extraction and ethanol precipitation. The following 59-32 P-labeled primers were used: betT-120R (59-GGCCGAGAAGTCGCGAAACA) and betT-273R (59-TGAATTCCGGTTTGGATTGT). RNA with labeled primers were annealed at 65uC for 15 min, slowly cooled down to room temperature for 2.5 h, and extended at 42uC for 1 h using AMV reverse transcriptase (New England Biolabs). The extended fragments were denatured at 95uC for 10 min and separated on 10% polyacrylamide gels containing 8 M urea.

Northern blot analysis
Forty micrograms of the total RNA sample prepared as describe above was denatured at 65uC for 10 min in twice the volume of formamide loading buffer and separated by electrophoresis on a 1.2% GTG agarose gel containing 2.3% formaldehyde. The gels were transferred onto nylon membranes (Hybond-XL, GE Healthcare) in 206 standard saline citrate. The random hexamer probes used for betT mRNA detection were synthesized using a random-primed DNA labeling kit (Roche; Pleasanton, CA), in which the PCR products containing the coding region of betT were used as a template. The primers used were betT-F and betT-R. The oligonucleotide probe used for M1 was rnpB-137R (59-GCTCTCTGTTGCACTGGTCG).
b-Galactosidase assays b-Galactosidase activity in whole cells was determined as previously described [24].

RNase III negatively regulates the expression of betT
To investigate whether the absence of RNase III affects the abundance of betT mRNA, we measured the steady-state levels of betT mRNA in wild-type and rnc-mutant E. coli by using semiquantitative RT-PCR. Two sets of primers were used to amplify cDNAs (+1 to +652 and +1612 to +2034) ( Figure 1A). Consistent with the microarray data from our previous study [7], rnc-mutant cells showed a 2.1-2.4-fold increase in the amount of betT mRNA compared to that observed in wild-type cells ( Figure 1A). To further explore the effects of RNase III concentration on betT mRNA decay, the half-life of betT mRNA was evaluated by a northern blot analysis ( Figure 1B). The half-life of the betT mRNA expressed in rnc-mutant cells was determined to be two-fold higher than that expressed in wild-type cells. Analogous results were obtained when the half-life of betT mRNA was measured by a semi-quantitative RT-PCR analysis ( Figure S1). For these experiments, M1 RNA, the 377 nt catalytic component of the t-RNA processing ribozyme RNase P, was used as an internal control because it is a stable RNA whose half-lives are 50-60 min in the exponential phase of growth [25]. These results indicate that both the steady-state level and half-life of betT mRNA correlate with the cellular RNase III concentration, indicating the involvement of RNase III in betT mRNA decay.

Identification of RNase III cleavage sites in betT mRNA
To investigate whether betT mRNA contains cis-acting elements that are responsive to RNase III, we performed an in vitro cleavage assay using purified RNase III and synthetic full-length betT mRNA transcripts. RNase III cleavage of 59-end labeled betT mRNA generated a major cleavage product that was approximately 80 nt in length and several minor cleavage products whose lengths were approximately 70-130 nt (Figure 2A). This result indicates that RNase III recognizes and cleaves specific regions of betT mRNA.
In order to identify the RNase III cleavage sites in betT mRNA, we performed primer extension analysis using a 59-end-labeled primer (betT+273R) that was designed to bind to a downstream region of the RNase III cleavage sites that were deduced from the in vitro cleavage assay. Because betT expression is induced at high salt concentrations, total RNA was purified from wild-type and rncmutant cells grown in plain (0.17 M NaCl) or 0.50 M NaClsupplemented Luria-Bertani media. However, we were not able to observe cDNA bands extended from betT mRNA in all reaction mixtures ( Figure S2, lanes [5][6][7][8]. This result indicated that the expression levels of betT mRNA were not high enough to be detected using primer extension analysis. Thus, the reaction was performed with the total RNA prepared from E. coli cells overexpressing betT mRNA and RNase III, and we observed several cDNA bands extending from betT mRNA that appeared to be RNase III-dependent ( Figure S2, lane 10). These bands were not present in the lane loaded with the reaction mixture containing total RNA extracted from cells that overexpressed betT mRNA in the absence of RNase III expression. To obtain a higher resolution of the cDNA bands, we used another 59-end-labeled primer (betT+120R) for primer extension analysis. We observed four distinct cDNA bands that were dependent on RNase III ( Figure 2B, lane 5 vs. lane 6). These bands corresponded to sites that are located in the double-stranded region of the betT mRNA coding sequence, and the cleavage of these sites by RNase III was predicted to produce cleavage products with a two-nucleotide overhang at the 39-end, which is a property of RNase III cleavage. These putative tandem RNase III cleavage sites were designated as sites I-IV ( Figure 2C). Four additional minor cDNA bands were also observed within close proximity of sites III and IV ( Figure 2B).
To biochemically demonstrate the cleavage of betT mRNA by RNase III, an in vitro cleavage assay was performed using a model hairpin RNA of betT mRNA containing RNase III cleavage sites I-IV in betT mRNA ( Figure 2C). In vitro RNase III cleavage of a 39-end-labeled model hairpin RNA generated four major cleavage products. The lengths of the major products corresponded to cleavage sites identified from primer extension analyses ( Figure 2D). We also detected other minor cleavage products that were likely produced by the random cleavage of transcripts by RNase III, which is an intrinsic property of RNase III in vitro [7,8,26]. We observed that the cleavage products at sites III and IV accumulated at similar rates, indicating that the RNase III cleavage at two tandem sites (I and IV; II and III) was independent ( Figure 2D).

RNase III cleavage determines betT mRNA stability in vivo
To test whether RNase III activity on cleavage sites regulated the stability of betT mRNA, we introduced a nucleotide substitution on cleavage site I or II (C33U or C39U) in the betT overexpression plasmid (pBetRS1) ( Figure 3A). These mutations did not alter the secondary structure of betT mRNA or the subsequent amino acid sequence. Total RNA was isolated from the MG1655rncbetT-harboring pRNC1 plasmid and either pBetRS1 or a derivative of pBetRS1 (pBetRS1-MT), and primer extension analysis was performed. Whereas the cDNA bands corresponding to RNase III cleavage sites I-IV were clearly visible in the reaction mixture containing wild-type betT mRNA, the abundance and patterns of cDNA products from mutant betT mRNAs differed from those of wild-type betT mRNA ( Figure 3B). Nucleotide substitution at site II (C39U) dramatically reduced RNase III cleavage activity at sites II and III, whereas the C33U mutation resulted in inhibition of RNase III cleavage at the mutated site and alterations of cleavage patterns. Despite the differences in the effects of these cleavage site mutations on RNase III activity on betT mRNA, the ratios of the total intensity of the cDNA bands corresponding to the RNase III cleavages sites to those corresponding to the transcriptional initiation site (TIS) were greatly decreased when these mutations were introduced (5.0-fold and 4.7-fold decrease for the C33U and C39U mutation compared with that of the wild-type, respectively), resulting in the significant accumulation of uncleaved mutant betT mRNAs. RT-PCR analysis of betT mRNA further confirmed increased steady-state levels of betT mRNA resulting from the mutations ( Figure 3C). These results suggest that the inactivation of either dimeric cleavage site (I and IV or II and III), which is handled by a single processing center formed by dimeric RNase III [4], led to an increase in the in vivo stability of betT mRNA.

Osmoregulation of betT expression by RNase III
Previous studies have demonstrated that RNase III activity is regulated by osmolarity within E. coli cells [7] and that BetT plays functional roles in the regulation of cellular osmolarity under low choline conditions [18,27,28]. These observations led us to investigate whether reduced RNase III activity caused by high osmolarity increases the stability of betT mRNA. First, we confirmed the functional role of betT in E. coli growth by culturing cells under hypo-osmotic (0.01 M NaCl), normal (0.17 M NaCl), and hyper-osmotic (0.50 M NaCl) conditions in the presence of low choline concentration (10 mM) ( Figure 4A and S3). The betTdeleted E. coli cells showed reduced growth rates compared to those of wild-type cells under hyper-osmotic conditions, indicating the importance of betT expression under hyper-osmotic stress with low choline concentration as previously shown [16]. The growth of proV-deleted bacterial cells was comparable to that of wild-type cells as previously reported [27][28][29][30]; the ProU transport system is involved in the osmotic stress response under high choline conditions. Next, we measured the half-lives of betT mRNA under different osmotic stress conditions. The results from semi-quantitative RT-PCR analysis showed that the decay rate of betT mRNA in wildtype E. coli under the hyper-osmotic condition was approximately two times lower than those under normal or hypo-osmotic conditions ( Figure 4B, left panel). Steady-state levels of betT  The 59-end-labeled betT transcript (4 pmol) was incubated with purified RNase III (1 pmol) in cleavage buffer with (+) or without (-) MgCl 2 at 37uC. The size of the cleavage products was estimated using size markers generated by internally labeled transcripts. The major cleavage products are indicated with arrows. Other minor cleavage products are indicated with asterisks. (B) Primer extension analysis of betT mRNA. Total RNA was prepared from MG1655rnc harboring pBetRS1 and either pKAN6B or pRNC1, which exogenously overexpressed betT mRNA, hybridized with a 59-32 Pend-labeled primer (betT+120R), and extended using AMV reverse transcriptase. Sequencing ladders were produced using the same primer used in cDNA synthesis and PCR DNA, encompassing the betT gene as a template. (C) The predicted secondary structure of betT mRNA region encompassing RNase III cleavage sites. The secondary structure was determined using the M-fold program [33]. (D) In vitro cleavage of the model hairpin RNA of betT mRNA. 39-end-labeled betT model hairpin (25 pmol) was incubated with purified RNase III (1 pmol) in a cleavage buffer with (+) or without (2) MgCl 2 , respectively. Cleavage products (I, II, III, and IV) were identified using size markers generated by alkaline hydrolysis and RNase T1 digestion. Relative abundance of each cleavage product was assessed by measuring the radioactivity of each band and plotted. doi:10.1371/journal.pone.0100520.g002 mRNA were also approximately two times higher under the hyper-osmotic condition, showing a correlation between the halflife and abundance of betT mRNA. The decay rates of betT mRNA in rnc-mutant cells under normal and hyper-osmotic conditions were similar with that in in wild-type E. coli under the hyperosmotic condition ( Figure 4B, right panel), indicating that decreased RNase III activity is associated with a prolonged halflife of betT mRNA in wild-type E. coli under the hyper-osmotic condition. Steady-state levels of betT mRNA in rnc-mutant cells were not significantly changed, indicating that transcriptional activation of betT mRNA expression did not occur under the hyper-osmotic condition that used in this study ( Figure 4B). We also observed that betT mRNA decays faster, especially between 2 min and 4 min after inhibiting transcription by the addition of rifampicin to the cultures, in rnc-mutant cells under the hypo-osmotic condition ( Figure 4B, right panel). Although further study is needed to explain this phenomenon, we think that betT mRNA decays via an unknown pathway in the absence of RNase III activity under hypo-osmotic condition. Nonetheless, these results indicate that increased steady-state levels of betT mRNA are mainly associated with decreased RNase III activity under conditions of hyper-osmotic stress. We further measured the half-lives of betT mRNA in strains that express betT mRNA from an osmotic stress-insensitive, constitutive promoter. MG1655betT harboring pBetRS1 was used for these experiments. The half-life of betT mRNA in MG1655betT harboring pBetRS1 under the hyper-osmotic condition was approximately 1.5 times higher than those under normal or hypo-osmotic conditions (5.1 min vs. 7.6 min) ( Figure 4C). Steady-state levels of betT mRNA were also approximately 1.7 times higher under the hyper-osmotic condi-  Figure 2B. (C) Effects of an RNase III cleavage site mutation on steady-state levels of betT mRNA determined by semi-quantitative RT-PCR analysis. Total RNA was prepared from strains MG1655rncbetT harboring pRNC1 and either pBetRS1 or its derivative (pBetRS1-C33U or pBetRS1-C39U), which were grown in LB at 37uC to an OD 600 of 0.6. doi:10.1371/journal.pone.0100520.g003 Total RNA was isolated from the cultures of MG1655 and MG1655rnc grown under the same conditions as described above, except that they were grown until an OD 600 of 0.6 was reached followed by the addition of rifampicin (1 mg ml 21 ). (C) Effects of the betT promoter on the half-life of betT mRNA. MG1655betT harboring pBetRS1 was grown as described above. Semi-quantitative RT-PCR analysis was performed to measure the relative amounts of betT mRNA. Experiments were performed in triplicate and repeated at least three times. doi:10.1371/journal.pone.0100520.g004 tion, once again showing a correlation between the half-life and abundance of betT mRNA. These results indicate that the betT mRNA decay is controlled by RNase III in response to osmotic stress. We confirmed that RNase III activity is downregulated under hyper-osmotic conditions using the E. coli strains KSC004 and KSC006. These E. coli strains contain an RNase III target site in single copy of a pnp9-9lacZ (KSC004) or an rnc9-9lacZ (KSC006) reporter gene [31,32]. Following a hyper-osmotic shift with 0.50 M NaCl, the RNase III activity in the KSC004 strain decreased by approximately 2.8-fold relative to cells that were under hypo-osmotic (0.01 M NaCl) or normal (0.17 M NaCl) conditions ( Figure S4). Analogous results were obtained with the KSC006 strain. The expression levels of RNase III protein did not significantly change under different osmotic conditions ( Figure S4), which is in agreement with the study conducted by Sim et al. [7].
Our results indicate that RNase III activity under hyper-osmotic stress conditions was downregulated, which is associated with an increase in the stability of betT mRNA.

Discussion
Although the factors involved in osmoregulation mechanisms in E. coli have been well studied, the regulation mechanisms underlying rapid control of osmosensing factors are not fully understood. Recent studies have demonstrated that RNase III plays roles in the regulation of several stress response factors, such as the biofilm-dependent modulation protein (Bdm) [7], osmosensing transporter protein (ProU) [9], and cobalt resistance protein (CorA) [8], at the post-transcriptional levels in E. coli. The present study showed that under conditions of hyper-osmotic stress, betT expression is also regulated by RNase III at the post-transcriptional level.
Given that RNase III enzymes require a minimum of a 22-bp stem for cleavage activity, RNase III cleavage at one dimeric site can abolish RNase III cleavage activity at the other dimeric site in betT mRNA. The results of the in vitro cleavage assay support this notion as the cleavage products of the 39-32 P-end-labeled model hairpin RNA that were generated by RNase III cleavage at sites proximal to the 39-end of the transcript (III and IV) appear to accumulate at similar rates, indicating that the RNase III cleavage at the two tandem sites (I and IV; II and III) is independent ( Figure 2D). In addition, the primer extension analysis on betT mRNA also indicated a correlation in the intensity of the cDNA bands between the two cleavage sites (I and IV; II and III). These results demonstrate that the RNA fragments generated by RNase III cleavage at one dimeric RNase III site in betT mRNA are not likely to be further cleaved by RNase III, indicating the independent cleavage of two tandem RNase III sites in betT mRNA. This result raises the question as to why betT mRNA contains two consecutive RNase III cleavage sites, neither of which is sufficient on its own for efficient betT degradation. This appears to be a unique property, as no other known RNase III substrates show two tandem RNase III cleavage sites. One possibility is that both sites are cryptic and do not provide the structural determinants that are required for full RNase III activity. This possiblity is supported by our mutational analyses at the betT mRNA cleavage sites, which demonstrated that inactivation of RNase III cleavage at either site (I and IV or II and III) was sufficient to inhibit betT mRNA degradation (Figure 3). Similar to two other RNase III-controlled mRNA species that encode factors responsive to osmotic stress, our findings show that steady-state levels of betT mRNA are dependent on RNase III activity under conditions of osmotic stress ( Figure 4C). The correlation between the increased half-life of betT mRNA and decreased RNase III activity under hyper-osmotic conditions ( Figure 4) further implicates the osmoregulation of RNase III activity on betT mRNA. It has been reported that RNase III activity can be regulated in E. coli under other conditions such as bacteriophage infection and cold-or aminoglycoside antibioticstress [11,32,34]. Protein regulators for RNase III have been also identified: Bacteriophage T7 protein kinase activates RNase III by phosphorylating the enzyme on serine [34] and YmdB inhibits dimerization of the enzyme [32]. However, these protein regulators do not appear to be involved osmoregulation of RNase III activity [7]. An unknown signal may be triggered by exposure of E. coli to hyper-osmotic stress that inactivates RNase III, thereby inhibiting the digestion of betT mRNA, resulting in increased cellular expression of BetT. Although the exact mechanism for the downregulation of RNase III activity under hyper-osmotic stress is unknown, we speculate that there is an RNase III-mediated osmoregulatory mechanism by which the expression levels of osmosensing factors, including Bdm, ProU, and BetT, are rapidly balanced out via the modulation of RNase III activity upon exposure to osmotic stress. Figure S1 Effects of RNase III on the half-life of betT mRNA. Total RNA was isolated from the cultures of MG1655 and MG1655rnc grown at 37uC in LB medium until an OD 600 of 0.6 was reached followed by the addition of rifampicin (1 mg ml 21 ). Semi-quantitative RT-PCR analysis of the cDNA +1-+652 was performed to measure the relative amounts of betT mRNA by setting the amount of betT mRNA in cultures 0 min after the addition of rifampicin to 1. The experiments were repeated three times and averaged. The error bars (standard errors of the mean) were used to indicate the range of assay results. The half-lives of betT mRNA were estimated by fitting and extrapolating the plots in the graphs. (EPS) Figure S2 Primer extension analysis of betT mRNA. Total RNA was prepared from MG1655, MG1655rnc, MG1655rnc harboring pBetRS1, and either pKAN6B or pRNC1. Total RNA was hybridized with a 59-32 P-end-labeled primer (betT+273R) and extended using AMV reverse transcriptase. Sequencing ladders were produced using the same primer used in the PCR DNA, encompassing the betT gene as a template.